All cancer screening tests that have had a significant impact on disease morbidity and mortality are aimed not at detecting early stage cancers, but pre-cancerous or pre-invasive lesions. Searching for the precursor lesion of ovarian epithelial tumors implies knowledge of their actual site of origin. Although the currently favored theory states that ovarian epithelial tumors arise from the mesothelial cell layer that lines the ovarian surface (ovarian coelomic epithelium), this theory does not account for the well-known fact that these tumors show features of tissues embryologically derived from the mullerian ducts and for the fact that preneoplastic lesions found in surgical specimens of women undergoing prophylactic surgeries for familial ovarian cancer predisposition are most often located in the fimbriae of the fallopian tubes. This proposal is based on Dr. Dubeau's longstanding hypothesis that ovarian carcinomas do not arise from the ovarian coelomic epithelium, but from tissues that are embryologically derived from the mullerian ducts. Aim #1 will use transgenic mouse models to directly determine the exact contribution of the mullerian ducts to the adult female reproductive tract and test the hypothesis that the ovarian surface epithelium is not of mullerian origin while specific para-ovarian and para-uterine structures such as rete ovarii and others have such an origin. The possibility that portions of the coelomic epithelium can re-program their differentiation state to become mullerian-like, which is an important component of the hypothesis that ovarian tumors are of coelomic origin, will also be investigated. Aim #2 will take advantage of a mouse model recently developed in Dr. Dubeau's laboratory that suggests that inactivation of Brca1 in ovarian granulosa cells stimulate neoplastic transformation in tissues that are embryologically derived from the mullerian ducts. We will use technologies similar to those used in aim #1 to determine whether the tumors that develop in these mice are of mullerian or coelomic origin. We will also knockout the p53 and Brca1 genes either in derivatives of the mullerian ducts or in coelomic epithelium and will cross the resulting mouse lines with our line carrying a knockout of Brca1 in granulosa cells. We anticipate that the tumors that will develop in these double mutants will have increased malignant potential and we will test the hypothesis that tumors in mice in which the additional mutations are targeted to the mullerian tract will most closely resemble human tumors morphologically and immunohistochemically. We will determine which portions of either the normal mullerian tract or coelomic cavity shows the most resemblance to those tumors based on their overall gene expression profile and the manner in which they respond to different phases of the estrus cycle, the equivalent of the human menstrual cycle. We will also test the hypothesis that tumors located in the para-uterine region are molecularly similar to those arising in the oviductal/ovarian region but that their anatomical location determines the likelihood that they will be cystic, which is a frequent characteristic of human ovarian tumors.